Abstract
OBJECTIVES: To compare the risk of pneumonia in older adults receiving donepezil, galantamine, or rivastigmine for dementia. DESIGN: Retrospective cohort study. SETTING: Nationally representative 5% sample of Medicare databases. PARTICIPANTS: Medicare beneficiaries aged 65 and older who newly initiated cholinesterase inhibitor therapy between 2006 and 2009. MEASUREMENTS: Pneumonia, defined as the presence of a diagnosis code for pneumonia as the primary diagnosis on an inpatient claim or on an emergency department claim followed by dispensing of appropriate antibiotics. Cox proportional hazards models were used to estimate the risk of pneumonia. Subgroup analyses and sensitivity analyses were conducted using alternative pneumonia definitions and adjustments using high-dimensional propensity scores to test the robustness of the results. RESULTS: The mean age of 35,570 new users of cholinesterase inhibitors (30,174 users of donepezil, 1,176 users of galantamine, 4,220 users of rivastigmine) was 82; 75% were women, and 82% were white. The cumulative incidence of pneumonia was 51.9 per 1,000 person-years. The risk of pneumonia for rivastigmine users was 24% lower than that of donepezil users (hazard ratio (HR)=0.75, 95% confidence interval (CI)=0.60-0.93). Risk in galantamine users (HR=0.87, 95% CI=0.62-1.23) was not significantly different from risk in donepezil users. Results of subgroup and sensitivity analyses were similar to the primary results. CONCLUSION: The risk of pneumonia was lower in individuals receiving rivastigmine than in those receiving donepezil. Additional studies are needed to confirm the findings of pneumonia risk between the oral and transdermal forms of rivastigmine and in users of galantamine.